不同任务类型下查询重构行为分析

Analysis of Query Reformulation Behavior in Different Task Types

以高效用查询重构推荐为导向,从查询重构类型入手,揭示各重构类型应用的频率、情境和有效性,以及任务对其的影响,以期提高查询推荐的准确性。101个调查对象共完成606个查询任务,其中222个任务包括查询重构行为,共计420次查询重构行为,以此为对象进行分析。结果显示:①事实型和信息扩展型任务中,具体化应用最多,决策型任务中,则是词语替换最多;②信息扩展型任务中,用户发现有用信息后,更多使用一般化查询重构策略;③事实型任务中,具体化有效性较高,信息扩展型任务中,具体化和词语替换具有较高有效性,而决策型任务中,则是一般化和词语替换相对有效。系统可以据此为用户提供有针对性的、高效用的查询提示服务,提高查询效果和用户满意度。

In order to recommend high utility queries, that is, to improve query recommendation accuracy, the frequency, context, and effectiveness of query reformulation types are revealed and how tasks impact them are explored . A total of 606 information searching tasks were completed by 101 respondents. 420 times query reformulations occurred in 222 tasks. Results showed that： （1）Specialization was most frequently used in Fact Finding and Information Gathering tasks, and Word Substitution was most frequently used in Decision Making tasks; （2）Generalization was more likely used after saving a useful web page in Information Gathering tasks; （3）Specialization was relatively more effective in Fact Finding tasks, Specialization and Word Substitution were more effective in Information Gathering tasks, and Generalization and Word Substitution were more effective query reformulation types in Decision Making tasks. According to these, the system can provide users with targeted and efficient query hints, which can improve the query results and user satisfaction.